基于深度自适应聚类算法的数据分层分类存储技术

Hierarchical Classification Storage of Data Based on Deep Adaptive Clustering Algorithm

为了提高电力数据信息管控能力,提出一种新型的数据分层分类存储方法。该方法采用深度自适应聚类算法,既可以有效识别单体数据的特征,还可以对数据整体结构进行一定程度的识别。通过自编码器DNN模型和图神经网络GCN模型,可以高效地实现复杂数据的分层分类。基于该算法的数据存储系统,可以将电力数据信息的多种数据信息转换为数字信息。通过深度学习算法分析数据的内部联系,提高系统的存储能力。实验结果表明,该方法对低数据量、复杂数据量和高数据量的分类准确率分别能达到97.5%、92%和86%,数据分类效率达到97%左右。

In order to improve the power data information management and control ability,a new data hierarchical classification storage method is proposed.This method adopts deep adaptive clustering algorithm,which can not only recognize the features of single data effectively,but also recognize the whole structure of data to a certain extent.By autoencoder DNN model and graph neural network GCN model,the hierarchical classification of complex data can be realized efficiently.The data storage system based on algorithm can convert various kinds of data information of power data information into digital information.The deep learning algorithm is used to analyze the internal connections of data,improve the storage capacity of the system.Experimental results show that the classification accuracies of low data volume,complex data volume and high data volume can reach 97.5%,92%and 86%,respectively.The data classification efficiency can reach about 97%.